Defining the surface areas of proteins that could sensitize individuals to an allergen or a viral antigen represents a major unsolved problem in food safety regulation and vaccine design. The overall goal of this project is to develop and validate novel computational methods to predict conformational epitopes and correlate the results with experimental data on observed cross-reactivity. In our current project period we maintained and further developed the Structural Database of Allergenic Proteins (SDAP), a database for analysis of viral sequences and structures, Flavitrack, and implemented motif search methods for linear epitopes. In this renewal project we will take these bioinformatics tools to the next level by including 3D structures of protein antigens. Our first aim is to develop and validate a novel data-mining tool for finding similar conformational epitopes in 3D databases. The scoring function uses pair and triple correlation data of our 5 quantitative descriptors derived from structurally characterized antibody-antigen interactions. The new tool will then be combined with our linear motif search method, based on physical-chemical property (PCP) motifs and property distance (PD) values to find protein antigens that are potentially cross-reactive. Once validated the new 3D search tool can also be implemented in other 3D-databases and supplement similarity searches that use only sequence information. Aim 2 is to create and validate a new flexible docking method for antigen-antibody complexes using self-correcting distance geometry to detect inconsistent constraints. Virtual screening methods will be developed to find similar interfaces of related proteins with known 3D-structures of components of protein complexes. The predicted interface regions from the database search will provide the geometric constraints, which will be used in a program similar to those we previously developed for interpretation of NMR data and for homology modeling. In Aim 3, the methods will be applied to two distinct, important ongoing projects: defining areas of clinically relevant, cross-reactive allergens that could be involved in IgE binding, and in designing multivalent vaccines. Our project will contribute to quantitatively estimating the potential allergenicity of cross-reactive antigens, and of novel proteins introduced into food products via genetically engineered organisms. We will archive data on clinically observed cross-reactivity from the literature in a machine-readable form and compare the observations with our results from linear motif search, 3D screening methods and the self-correcting distance geometry calculations on antibody-antigen complexes. Our new software tools can also guide the design of multivalent vaccines against flaviviruses. Many mutations are known that alter the resistance of these viruses to neutralizing antibodies. Guided by this specific data, our tools will improve the presentation of divergent epitope areas, so that they can generate a more effective protective antibody response.